HINS: HLA-mediated immunogenic neoantigen score for robust prediction of immune checkpoint blockade response

Abstract

Immune checkpoint blockade (ICB) therapy response prediction provides remarkable genomic and transcriptomic gains and has successfully understood the molecular mechanisms underlying various cancers. However, only a subset of patients with advanced tumors currently benefits from ICB therapies, which at times incur considerable side effects and costs. Developing predictive tools for ICB response has remained a serious challenge because of the complexity of the immune response and the shortage of large cohorts of ICB-treated patients that include both omics and response data. Here, based on a pooled genomic dataset of 919 patients across multiple solid tumors treated with anti-PD-(L)1 or anti-CTLA-4, we constructed human leukocyte antigen class I (HLA-I)-mediated immunogenic neoantigen score (HINS). HINS is a predictor of ICB response using neoantigen quality and HLA evolutionary divergence to integrate the factors associated with immune activation and evasion. It yielded an overall accuracy of AUC=0.853, outperforming existing predictors and capturing more than 75% true responders. Experimental analysis indicated that patients with higher HINS were more likely to undergo survival benefits following ICB therapy. Our results highlighted the transcriptional and genomic correlations between HINS-identified molecule features and ICB response, such as immunoresponsive gene enrich pathways, favorable and unfavorable genomic subgroups, and hotspot somatic events in driver genes. This study presented an interpretable, accurate deep-learning method using meaningful predictive features to predict the response to ICB, providing biological insights into the complexity of the determinants underlying immunotherapy.